Content Replication Strategy in BitTorrent File Sharing System

Abstract

Peer-to-Peer (P2P) file sharing systems, most notably BitTorrent (BT), have achieved tremendous success among Internet users and played an important part in quick prosperity of Internet. However, the accompanying huge backbone traffic has brought numerous Internet Service Providers (ISPs) close to break-point. How to coordinate the conflict between ISPs and BT systems has always been the center of attention in the research of worldwide network architecture, for which many institutions had supposed a great deal of solutions. But new research result of network measurement shows that locality policy of peer selection may not be the best choice. In this paper, we focus on the downloadable file content resources scheduling, examine different peer properties and BT swarms distribution in Autonomous Systems (ASes) through large-scale measurement, replicate the most desirable torrent file content to designated BT peers, and keep the total file resources balance of BT network. In this way, most BT users could download the desired file locally and naturally, and the inter-ISP traffic would be reduced substantially. Our trace-driven simulation shows that outside traffic is decreased by 50% to 70%.